Topic: BFL board project COINTAMINATION - EU facility - ORDERS OPENED for Chili - page 16. (Read 73081 times)

I will go now too: for me this problem have big possibilty to be fixed by firmware and have some limited rework capacity if found later problem is HW related

Regards

what will be the difference between asking for delivery now, ..........and waiting a few more days for problem to be fixed?

You are right about small components. I also asked to be investigated... Will asked again when they make my broads as they are.

If anyone would like them too now is the time to say.

This kind of problems is really frustrating: in older project where worked I lost one week to realize people have changed 2 reel position on pick and place machine. Small components usually are not marked so can be difficult to spot changes.
Also from my experience problems arinsing many days or weeks after normal work usually are eletrolitic capacitor related ( more if heat are involved ).

The "black shadow" of BFL delays seems to darken the path of this project but please don’t give up so close to end. I agree at this moment ROI probably would never achieved but people are in  this area not only for money. Keep you good work  

I did test my boards with 7 different PSU… From cheap 20€ 450W to hi end 350€ 1200W… No differences…

Problems w/ 12V supply to board are out of question?

Regards

My instinct goes for the power controller, (but maybe I am Influenced by the failure on my chili)

Thank you for your detailed explanation of what is happening or has happened. I actually was not looking for that. All I did was realize today (yesterday) that the original Cointamination board would have taken the same amount of time to get fixed/running. I have written off the investment I made in this project 5 weeks ago (1 weeks after I send my chips to you) and I realized the project (at least for me) will never ROI (in BTC or FIAT). All I want is working boards 39+ GH/s with no special tweaks to get them running in the first place, otherwise the resale value would be ZERO ... which makes is even worse ...  I know the chips are not the problem, something else is. I honestly don't care anymore how long it takes to get the boards working. I equally admire your dedication (to this project) to get everything running!! So, DO NOT GIVE UP!!! You will find the problem and be able to get rid of it! And once I have my boards, I still will have great joy in build my crazy water-cooled miner, because I love to build things.

All the best!

     one4many

X-ray checked... Unless the problem is chip itself...

Could it be the chips or the way they are being mounted that is the problem?

Probably yes... Unless you are ready to have it made as it is... If data I have are correct and no part is defective (and we can't detect defect) I'm 99% sure it is put together right and it is just firmware reacting a bit different do to small differences since you never get 100% same board with different manufacturers(I'm told by ChipGeak that this is not the case but it is hardware error in 1V power supply but still it fits best). I will have my made(if I can't figure it out by the end of the week) as they are and will run them from more computer so it will be easier to make them run. Since they crash only at some specific speeds once you have them run it is go time...

What I plan to test(but this is really the last idea I have) is replace all 1V power components with one bought at later date. I also asked MrTeal to give me the list of them and to look at PnP data if there is a error for one of them. I also asked if there is a way I can get slightly changed firmware to test my theory of voltage/hashing speed problem...

We did look at MrTeal made board and compared it to ours. We did recheck parts used. We did look at all boards with IR and scope. Test components. Look for any spikes in voltage. Look for cold and bad connections. Looked how MrTeal board act and how my... The only think I can't do but this is also the case for MrTeal made board is communicate with 1V power controller. It crashing the board after some time if I do that. So not sure why... All this was not made on every single board since that is a lot of work but since all have same problem...

If anyone has any idea I didn't think of please let me know...

And interesting enough. Now that I used MrTeal made board for some time the board did crash the same way as my do.

Ouch.

That means ... Realistically we won't see our boards until the beginning of next year. 

No. We will see what will happen when MrTeal get his board...

So are you saying we still don't know what is wrong with the boards?  Has Mr. Teal made any progress?

Unfortunately only one board do this... And when running at full board get much hotter...

I did notice that marking on one of the components is a bit different then on a sample board but component datasheet say it is only a date code...

OK found something but not sure what to think... But it might be that I have a batch of defective components... Will need to look at all made boards...

This is one of a better running boards so...

https://www.dropbox.com/s/wvxuvls6502pdov/temps.jpg

Problem is that this was detected on this one but not on other we looked at... Tomorrow will go over all. It was late and I was asked to live friend lab...

I don't think it is as easy as that. It is a speed and voltage combination. If this is off the problem will not appear... So if you stop CGminer and voltage will not drop to 0,85V(or below crash voltage for board not sure) and restart CGminer it the board will not crash. Board made by facility did have 25% less copper on outer layers. But we did boards with found PCB to and same problem...

But yes I agree to a point. One normally working board is probably below problematic hashing/voltage speed. So there is probably a strongest link...

EDIT: If MrTeal agrees I can also give you tracking data for board send to him...

So the strongest board is the one that resets?

I imagine there's a correlation.  Like the board is running so hard it overwhelms some supporting piece.

I know the Avalon boards had an issue with voltage drop due to the PCB copper being too thin.  I wonder if something similar is happening here.

MrTeal thanks for all your help!